UPDATE: Since this is a security patch, I’ll be updating to a newer kernel anytime soon and will stay away from 3.2.29. However, since it would require more work, I’ll stay in 3.2.29 a bit until I have enough time to install any of the newer kernels (perhaps this coming weekend).

UPDATE: An updated dated June 3, 2013 was uploaded to Slackware mirrors and promised to fix the Intel issues with kernel 3.2.45 (3rd rebuild). Today, June 9, 2013, I have applied the patch and so far it works. I have uploaded some screenshot showing some Firefox 3d stuff.

Firefox 3d view is happy with the latest kernel rebuild!

Original post below.

Kernel patch 3.2.45

The original kernel for 14.0 was 3.2.29 and the supposed to be security patch kernel version is 3.2.45. After few days since the patch has been released, I’ve applied the patch via Slackware update. The installation went through and I’ve even updated the initrd so I could use the generic kernel. However, during my final reboot so I could start working on the machine, it freezes up few minutes later after logging in into KDE.

I’ve tried different desktop environment (XFCE) and even bare bone window managers (blackbox, fluxbox) but same issue occurs. I suspect it was the graphics driver. My Intel Graphics 3000 may not be supported or whatsoever. I’ve read the change logs and it says that there was another update few days after the kernel patch. I’m sure I got the latest patch but I still experienced the issue described. There was even a discussion in the forum about the said issue.

Stuck due to no wifi

I’m using NetworkManager for my wifi. Since I’ve configured the wifi via NetworkManager, by default I could not connect to the internet without logging into X. I was planning to downgrade but couldn’t download due to no internet connection in runlevel 3. I’ve asked around again so I could do something to save my machine.

nomodeset

I’ve read somewhere on that forum post that we could pass nomodeset parameter into lilo or any boot manager so it would bypass awesome graphics features and thus allow troubleshooting (ex: can connect to the internet but with crappy graphics quality). Since I’m using lilo, I do the following:

• Start the machine
• At boot screen, press tab
• Type in the name of the lilo entry for your linux, mine I have the following entries:
  • Linux
  • Linux3245
  • Windows
• After typing in the lilo entry, add space and type in nomodeset
  • Linux3245 nomodeset
• Press enter
• Wait for the login manager to show up and login to the desktop manager
• Once you are able to login, you can download and install the old kernel package

Downgrade kernel

Now that I’m able to login to the desktop manager and connect to the internet, I was able to search further about the issue. However, my plan is clear – to downgrade to the original 14.0 kernel version 3.2.29. Since I already have a full copy of Slackware 14.0 including the source, the original kernel is also available in that copy. I just installed old kernel in the slackware/a directory.

cd /path/to/slackware-14.0
cd slackware/a
installpkg kernel*

Now that I have both kernel 3.2.45 and 3.2.29, I can choose to use both and play around. But I choose to just revert the default kernel to 3.2.29 and leave it as is for the next couple of months. The task is simply to re-run the mkinitrd command, update lilo.conf to point to the correct generic kernel, rerun lilo and rebooted.

I know that messing around with Linux kernel is scary but it has been done and I’ve committed the same mistake over and over again. Until the next kernel trouble my friend. For now I could go back to work instead of dealing with OS problems.

UPDATE: I’ve just updated the post after discovering that the script can be further enhanced by using bash functions. See updated script at the bottom.

The project

The project is not that large, however, there are some 5 or 6 sites on it with multiple environments like development environments for multiple developers, staging and production. I need to run tests for all sites at specific environment. I’m using behat as tool for running automated tets (BDD). Below is the test directory structure.

|-- test_dir_root
    |-- site1
    |   |-- behat.yml
    |   `-- features
    |-- site2
    |   |-- behat.yml
    |   `-- features
    |-- site3
    |   |-- behat.yml
    |   `-- features
    |-- site4
    |   |-- behat.yml
    |   `-- features
    |-- site5
    |   |-- behat.yml
    |   `-- features
    `-- site6
    |   |-- behat.yml
    |   `-- features
    `-- run_tests.sh

Strategy

The strategy is simple, go to test’s root directory, loop through all directories (listed in an array) then run the tests with uniform parameters, ex: if I want to run a staging test, it should run tests against staging environment for all sites listed.

Since behat‘s default behavior is to run in verbose mode, I have to pass some parameters to have a less cluttered test output.

The Script

I’ve put script at the root directory for the tests and name it run_tests.sh. Below is the code.

#!/bin/sh

if [ "$1" == "" ]; then
    echo "Usage: ./run_tests.sh <production|staging|development>" 
    echo "Ex: ./run_tests.sh staging" 
    exit 1
fi

TEST_ROOT_DIR=$(pwd)
SITES=("site1" "site2" "site3" "site4" "site5" "site6")

echo "Initiating tests..."

for i in "${SITES[@]}"
do
    :
    # Go to test root dir
    cd "$TEST_ROOT_DIR"
    TARGET_DIR="$TEST_ROOT_DIR/$i"
    COMMAND="behat --profile $1 --format progress"
    cd "$TARGET_DIR"
    echo "Running tests for $i at $(pwd)"
    $COMMAND
    echo ""
done

As you can see, it has some basic control structure like if statement, loops and even string concatenation. The code block:

if [ "$1" == "" ]; then

checks if the first command line argument is empty. If you are using two command line arguments, it will be at $2, third will be at $3 and so on. The code block:

SITES=("site1" "site2" "site3" "site4" "site5" "site6")

declares an array. The array is then used for looping. The code block:

TARGET_DIR="$TEST_ROOT_DIR/$i"

concatenates the string variable $TEST_ROOT_DIR and $i with / at the middle.

Let’s run it. Assuming they all passes, below will be how it will look like.

lysender@darkstar-lenovo-32:/home/data/www/huge_project/behat$ ./run_tests.sh darkstar
Initiating tests...
Running tests for site1 at /home/data/www/huge_project/behat/site1
....................

2 scenarios (2 passed)
20 steps (20 passed)
0m0.211s

Running tests for site3 at /home/data/www/huge_project/behat/site3
..............................................................

18 scenarios (18 passed)
62 steps (62 passed)
0m1.238s

Running tests for site3 at /home/data/www/huge_project/behat/site3
....................

2 scenarios (2 passed)
20 steps (20 passed)
0m0.214s

Running tests for site4 at /home/data/www/huge_project/behat/site4
.........F.........F

(::) failed steps (::)

01. ...

02. ...

2 scenarios (2 failed)
20 steps (18 passed, 2 failed)
0m0.226s

Running tests for site5 at /home/data/www/huge_project/behat/site5
.........F.........F

(::) failed steps (::)

01. ...
02. ...

2 scenarios (2 failed)
20 steps (18 passed, 2 failed)
0m0.218s

Running tests for site6 at /home/data/www/huge_project/behat/site6
....................

2 scenarios (2 passed)
20 steps (20 passed)
0m0.214s

lysender@darkstar-lenovo-32:/home/data/www/huge_project/behat$ 

That’s it. Enjoy and share.

Update!

After being annoyed by running all tests for all sites every single time, I decided to add an option to I could just run a test for just one site. Therefore, it will either run tests for all sites or run test for just one site. Below is the updated script.

#!/bin/sh

if [ "$1" == "" ]; then
    echo "Usage: ./run_tests.sh <production|staging|development> <optional site_name>" 
    echo "Ex: ./run_tests.sh staging" 
    echo "Ex: ./run_tests.sh staging site2" 
    exit 1
fi

TARGET_SITE=""
if [ "$2" == "" ]; then
    SITES=("site1" "site2" "site3" "site4" "site5" "site6")
else
    SITES=("$2")
fi

BEHAT_PROFILE="$1"
TEST_ROOT_DIR=$(pwd)

run_behat_test_on_target() {
    # $1 is root dir
    # $2 is the target site
    # $3 is the profile name
    TARGET_DIR="$1/$2"
    COMMAND="behat --profile $3 --format progress"
    cd "$TARGET_DIR"
    echo "Running tests for $2 at $(pwd)"
    $COMMAND
    echo ""
}

echo "Initiating tests..."

for i in "${SITES[@]}"
do
    :
    run_behat_test_on_target $TEST_ROOT_DIR $i $BEHAT_PROFILE
done

To run the script, we doe something like below:

# Run tests for just site 2
./run_tests.sh staging site2
# Run tests for all sites
./run_tets.sh staging

My old code goes like this.

// To check
$("#some_checkbox").attr("checked", "checked").checkboxradio("refresh");

// To uncheck
$("#some_checkbox").removeAttr("checked").checkboxradio("refresh");

This method, for some reason does not work for me in jQuery mobile (simulated in a desktop browser). I saw a sample in docs and tried to use it and it works flawlessly.

// To check
$("#some_checkbox").prop("checked", true).checkboxradio("refresh");

// To uncheck
$("#some_checkbox").prop("checked", false).checkboxradio("refresh");

Note that this may require higher version of jQuery. However, using:

$("#some_select").attr("selected", "selected").selectmenu("refresh");

works for select/dropdown elements. Weird.

I haven’t investigated further but there must be something wrong with my old code. Too lazy to investigate though.

Here is the security advisory in nginx site. Below are the steps I did when applying updates.

Updating nginx

In Slackware (mine is 14.0), nginx is not shipped as stock software. You get the source code and get a SlackBuild script and build it on your own. My build script and custom configuration is still in the server, therefore, all I have to do is put the latest source code, pass the new version and upgrade the installed nginx package.

Get the SlackBuild script for nginx on SBo.

cd /path/to/build/script
wget http://nginx.org/download/nginx-1.4.1.tar.gz
VERSION=1.4.1 ./nginx.SlackBuild 
upgradepkg /tmp/nginx-1.4.1-i486-1_SBo.tgz

Now that we’ve upgraded nginx, we need to check the configurations if it is still intact. If we’ve verified that we are still good with the configuration, restart the nginx server.

/etc/rc.d/rc.nginx restart

Note: My server is a VPS hosting some websites and also servers as staging server.

Always watch out for security advisories out there and secure your servers.

In your settings.py, the default logging configuration may be configured for I don’t know, is it production ready? For Django starters, this will hide basic errors you may encounter. We are going to configure the LOGGING dict to have the required formatters, handlers and loggers. Below is my LOGGING section of settings.py.

LOGGING = {
    'version': 1,
    'disable_existing_loggers': False,
    'filters': {
        'require_debug_false': {
            '()': 'django.utils.log.RequireDebugFalse'
        }
    },
    'formatters': {
        'verbose': {
            'format': '%(levelname)s %(asctime)s %(module)s %(process)d %(thread)d %(message)s',
        },
        'simple': {
            'format': '%(levelname)s %(message)s',
        },
    },
    'handlers': {
        'mail_admins': {
            'level': 'ERROR',
            'filters': ['require_debug_false'],
            'class': 'django.utils.log.AdminEmailHandler'
        },
        'console': {
            'level': 'DEBUG',
            'class': 'logging.StreamHandler',
            'formatter': 'verbose',
        },
    },
    'loggers': {
        'django.request': {
            # 'handlers': ['mail_admins'],
            'handlers': ['console'],
            'level': 'ERROR',
            'propagate': True,
        },
    }
}

Please note that this is for development mode only and for those who run the build-in web server. For production purpose, change the settings for the appropriate logging.

I’ve found the tip from a GIST.

The shortcut is simple, select the text to convert and press the following keys.

For non-mac keyboards.

• CTRL+ K, then L – for converting into lowercase
• CTRL + K, then U – for converting into uppercase

Example, you have a text NOFOLLOW and you want to convert into nofollow. Select the text, press CTRL and K at the same time, then release K while still pressing CTRL then press L.

That’s it. Enjoy and share.

Install nginx

First, we grab a SlackBuild script for nginx via Slackbuilds.org. Extract it, then grab the latest source code for nginx (currently at 1.4.0).

wget http://slackbuilds.org/slackbuilds/14.0/network/nginx.tar.gz
tar xvzf nginx.tar.gz
cd nginx
wget http://nginx.org/download/nginx-1.4.0.tar.gz

Now that the build script and the nginx source code is ready, we need to modify the build script so that it will install the 1.4.0 version instead of whatever version is being hardcoded into it by the build script. In my case, the version used by the script is still at 1.2.6.

Edit nginx.SlackBuild and look for the text VERSION=${VERSION}:-1.2.6 and change it to version 1.4.0. Next, go down the codes and look for the text ./configure and check if --with-http_realip_module is present. If not, add it in the middle of the ./configure code block.

Next, ensure that --with-ipv6 is present as well. If not, add it to the ./configure code block. These two settings allow sniffing real IP address from proxy servers or CDN like CloudFlare. The other option allows the ipv6 syntax that we can use with HttpRealipModule. I needed it on my setup but you may not need it on yours. We’ll never know.

Now that we are finally ready building nginx, let’s build it now (run as root). It may take several minutes to complete.

./nginx.SlackBuild

Finally, we need to install it on our Slackware box. The package filename will be displayed by the build script after it finishes compiling nginx and it is usually put under /tmp directory. (Run as root).

installpkg /tmp/nginx-1.4.0-i486-1_SBo.tgz

Configure nginx

Now that we already have an nginx installed, let’s configure it to be our webserver. First, we set the daemon up and running and be sure we closed/disabled all other web server like Apache. If we have an Apache server running, let’s kill it and make sure it will never run again upon restart.

/etc/rc.d/rc.httpd stop
chmod -x /etc/rc.d/rc.httpd

Next, lets make nginx and php-fpm RC script enabled upon boot.

chmod +x /etc/rc.d/rc.php-fpm
chmod +x /etc/rc.d/rc.nginx

I usually use vhosts for everything, therefore, in this post, I’ll jumpt straight into making virtual hosts for nginx. Let’s start with our default virtual host which is also the localhost. I call my virtual host as darkstar.darkstar.net. The document root will be on /srv/www/htdocs/darkstar.darkstar.net.

Create the directory and put an index file /srv/www/htdocs/darkstar.darkstar.net/index.html and put some content like It works! or any content just for testing. For testing local servers, you may need to add a host file entry for darkstar.darkstar.net like below:

# /etc/hosts
# Other contents
# ...

127.0.0.1       darkstar.darkstar.net

Next, create an nginx conf file. Below is my conf file for /etc/nginx/nginx.conf.

worker_processes  1;

events {
    worker_connections  1024;
}
http {
    include       mime.types;
    default_type  application/octet-stream;

    sendfile        on;
    #tcp_nopush     on;

    #keepalive_timeout  0;
    keepalive_timeout  65;

    #gzip  on;

    server {
        listen       80;
        server_name  darkstar.darkstar.net;

        #charset koi8-r;

        error_page   500 502 503 504  /50x.html;
        location = /50x.html {
            root   html;
        }

        root /srv/www/htdocs/darkstar.darkstar.net;

        location / {
            root        /srv/www/htdocs/darkstar.darkstar.net;
            index       index.html index.htm;
        }

        access_log  /var/log/nginx/darkstar.darkstar.net-access_log combined;
        error_log   /var/log/nginx/darkstar.darkstar.net-error_log error;
    }

    # Include other config/sites
    include /etc/nginx/sites-enabled/*;
}

As you’ve noticed, we didn’t put any php related configuration here. We will skip that for now and let this vhost serve static files only. At the end of the config file, the line that says include /etc/nginx/sites-elabled/*; means that we can add more vhosts on that directory without messing around with the main config file. Create the directories for now and leave it with no content.

cd /etc/nginx        
mkdir sites-available
mkdir sites-enabled

Now we are ready for our first test. Run the nginx daemon and let’s see what will happen.

/etc/rc.d/rc.nginx start

Now visit http://darkstar.darkstar.net. You should see the text It works! on your browser.

Configure nginx + php-fpm

Now that nginx is ready, lets get PHP working. In this post, we will simply make the PHP run. Nothing fancy. We call our demo site myproject.darkstar.net. You must add this site in your local host file if this is a local virtual host, like what we did on darkstar.darkstar.net. Next, add a config file at /etc/nginx/sites-available/myproject.darkstar.net.conf with below content.

server {
    listen      80;
    server_name  myproject.darkstar.net;
    root        /srv/www/htdocs/myproject.darkstar.net;
    index       index.html index.htm index.php;

    location / {
        try_files $uri $uri/ /index.php$is_args$args;
    }

    location ~ \.php$ {
        fastcgi_pass    127.0.0.1:9000;
        fastcgi_param   SCRIPT_FILENAME $document_root$fastcgi_script_name;
        include         fastcgi_params;
        #fastcgi_param   PHP_VALUE "auto_prepend_file=/home/data/scripts/common_functions.php";
    }

    location ~ /\.ht {
        deny  all;
    }

    access_log  /var/log/nginx/myproject.darkstar.net-access_log combined;
    error_log   /var/log/nginx/myproject.darkstar.net-error_log error;
}

Then make sure you sym-link the config file to the sites-enabled directory.

cd /etc/nginx/sites-enabled
ln -s ../sites-available/myproject.darkstar.net.conf myproject.darkstar.net.conf

Next, make a sample PHP script at /srv/www/htdocs/myproject.darkstar.net/index.php with some basic PHP code, ex:

<?php phpinfo() ?>

Launch

Finally, let’s launch the site with a test page. Start php-fpm, restart nginx.

/etc/rc.d/rc.php-fpm start
/etc/rc.d/rc.nginx restart

Open the site http://myproject.darkstar.net/ and you should be a violet phpinfo() page. By the way, with above nginx config, it allows basic rewrite feature found on frameworks like Zend Framework, Kohana, CodeIgniter and even WordPress.

Enjoy and share.

I haven’t received any notification about this change. The Fair Usage Policy page didn’t mentioned anything about exact GB limit but according to some blog posts, it is 15 GB per month.

Without downloaded stuff, I could easily exhaust that 15GB in less than a month. This is due to my nature of work where I read a lot, connect to some servers and transfer stuff a lot. I may adjust my workflow as I cannot connect to my VPS all the time anymore.

On the other hand, I just extended my contract with them for 7 months for a promo wherein you pay P100 less for 7 months. I signed up immediately but now that I know this policy, I regret. I deserve better internet.

The month just changed from April to May but it seems that my 15GB cap has not reset. I know they sucks but they could have sucked less.

To rebuild or not?

nginx has a module called HttpRealipModule where it allows setting real IP address for a given source IP by using the standard header values passed by proxy servers. This will allow us to tell nginx to pickup CloudFlare’s CF-Connecting-IP header value and use it as the real IP address and will be used when logging to access logs.

This feature requires that nginx is built with --with-http-realip_module build option. My current version is nginx 1.4.0 and luckily it has enabled --with-http-realip_module by default.

Since the CloudFlare documentation includes ipv6 IP addresses, we also need to ensure that our nginx can handle ipv6. This requires --with-ipv6 option when building nginx. Unfortunately, my nginx is not built with ipv6 support. I am forced to rebuild it. Luckily, I’m using Slackbuilds.org’s nginx Slackbuild script. I just edited the script to pass --with-ipv6 and re-installed nginx.

HttpRealipModule

This is the guide I used when setting up the CloudFlare real ip sniff. I’ve created a file called /etc/nginx/cloudflare-real-ip.conf.

# Cloudflare

set_real_ip_from   204.93.240.0/24;
set_real_ip_from   204.93.177.0/24;
set_real_ip_from   199.27.128.0/21;
set_real_ip_from   173.245.48.0/20;
set_real_ip_from   103.21.244.0/22;
set_real_ip_from   103.22.200.0/22;
set_real_ip_from   103.31.4.0/22;
set_real_ip_from   141.101.64.0/18;
set_real_ip_from   108.162.192.0/18;
set_real_ip_from   190.93.240.0/20;
set_real_ip_from   188.114.96.0/20;  
set_real_ip_from   197.234.240.0/22;
set_real_ip_from   198.41.128.0/17;
set_real_ip_from   2400:cb00::/32;
set_real_ip_from   2606:4700::/32;
set_real_ip_from   2803:f800::/32;
set_real_ip_from   2405:b500::/32;
set_real_ip_from   2405:8100::/32;
real_ip_header     CF-Connecting-IP;

Next, for all sites in your server that is CloudFlare enabled, include this config by putting a line in your vhost configuration like below:

server {
    # Some config
    # ...
    # ...

    # CloudFlare config
    include /etc/nginx/cloudflare-real-ip.conf;
}

Restart nginx and what your access logs. It should start logging real IP address from now on. Be sure to update the list of IP addresses from time to time since CloudFlare may add/remove IP address in the future.

Note: We didn’t actually configured nginx to allow ipv6, but instead we just make sure that our config syntax will work and not throw error. Without the ipv6 option, our config above will fail since it will not be able to understand the ipv6 syntax. However, we didn’t specifically mentioned how to allow ipv6 connections to nginx. This requires another configuration that I am not interested as of this time.